Spectral Features of the Symbiotic Variable Star Ch Cygni in 2005 - 2006

SPECTRAL FEATURES OF THE SYMBIOTIC VARIABLE STAR CH CYGNI IN 2005 - 2006

Kye Hwa Yoo 1 and Tae Seog Yoon 2 1 Department of Science Education, Ewha Womans University, Seoul, 120-750, Korea E-mail: [email protected] 2 Department of Astronomy and Atmospheric Sciences, Kyungpook National University, Daegu, 702-701, Korea E-mail: [email protected] (Received June 12, 2009; Accepted July 28, 2009)

ABSTRACT

This article reports the spectral behavior of CH Cygni, using data obtained in October 2005 and June 2006. In these epochs, CH Cygni showed emission lines of H I, Fe II, [Fe II], [O III], [N II], [Ne III] and [S II]. Many of these lines were more enhanced since 2004. The underlying M-type spectrum was removed to get the intrinsic emission profile, and the resulting profiles were deconvoluted into several Gaussian components. Also, the radial velocities for all the lines that appeared in these spectra of CH Cygni were measured. The resultant lines were compared with each other and with those obtained in 2004; the findings are explained in terms of an accretion disk and jets. Key words : Symbiotic variable star CH Cygni, enhancement of spectral lines, accretion disk, jets

I. INTRODUCTION of the M-type giant appeared to have nearly the same strength as those in 2004. However, weaknesses of the Before 1963, CH Cygni was classified as an M7 III- M-type lines were observed in the outburst phases. type (Yamashita, 1967). And, the first eruption of In Paper I and Paper II (Yoo, 2007), detailed infor- CH Cygni was observed in 1963 (Faraggiana and Hack, mation on the profiles and radial velocities for the lines 1969). At that time, the spectra of CH Cygni showed of CH Cygni observed in April 2004 and October 2004 the blue continuum and emission lines of H I, He I and were presented, respectively. [Fe II] overlain on the M III-type spectrum. In 1963 - 2000, a symbiotic variable star CH Cygni had sev- In this Paper, the variations in the spectral lines eral outburst phases (Yamashita and Maehara, 1979; of the high-resolution spectra secured in October 2005 Mikolajewski et al., 1990; Eyres et al., 2002; Skopal et and June 2006 using the spectrograph of the Bohyun- al., 2002). In the eruption and quiescence phases, CH san Optical Astronomy Observatory (BOAO) are re- Cygni was observed on X-ray, UV, visible, IR and radio ported. In the second section, the observations and wavelengths (Yoo, 2006, Paper I, references in there). data reduction procedures are described. In the third While the broadest emission component of H I lines section, the behaviors of the spectral lines observed in was caused by bipolar radio jets (Paper I), Contini et the present epochs are detailed in comparison with the al. (2008) proposed the broadest emission component behaviors of those obtained in 2004. Subtraction of the of H I lines was formed from downstream of a fast re- spectrum of the underlying M-type giant is provided in verse shock. the fourth section. Gaussian deconvolutions for Hα, [O III], [Ne III], [N II], and [S II] and the radial veloc- Two models of CH Cygni have been proposed; one ity measurements are furnished in the fifth and sixth is the binary system with the orbital period of 5750- sections, respectively. In the last section, the line fea- day which consists of an M III and a white dwarf (Ya- tures are discussed in detail and a conclusion is arrived mashita and Maehara, 1979), and the other is the triple at. system with orbital periods of 756-day and 5300-day which consists of an M III, a white dwarf and an un- II. OBSERVATIONS AND DATA REDUC- seen object (Hinkle et al., 1993). TIONS In 2004, CH Cygni had m ∼ 8.1 in the V band, whereas the magnitudes of CH Cygni in the V band Observations were carried out on October 13, 2005, in October 2005 and June 2006 were m ∼ 7.9 and ∼ and June 4, 2006, with a high resolution Echelle 8.2, respectively (AAVSO). In the period between Oc- spectrograph, BOES (BOAO Echelle Spectrograph), tober 2005 and June 2006 (called ‘the present epochs’ mounted on an 1.8 m reflector at the BOAO. A hereafter), the TiO bands and metal absorption lines 2048 × 4096 pixel CCD camera, with a size of 15 × 15 µm per pixel, was used. The spectral ranges used were Corresponding Author: K.-H. Yoo from 3800 - 7000 A.˚ A cross disperser divides the spec-

– 93 – 94 YOO & YOON

Table 1. Journal of observations of CH Cygni

File JD Date Phase Exposure Diameter of Weather number 2450000 UT φ time(sec) ﬁber(µm) condition es16335 3657.013 Oct. 13, 2005 0.52 1000 200 poor es22773 3892.096 June 4, 2006 0.57 1280 300 poor Phases are cited from the ephemeris of Hinkle et al.(1993).

7 20 CH Cygni CH Cygni 6 15 5 Resulting Resulting 4 10 3 2 5 CH Cygni Oct. 13 2005 relative intensity relative intensity CH Cygni Oct. 13 2005 1 0 M6 III (RZ Ari) 0 M6 III (RZ Ari) 6540 6550 6560 6570 6580 4840 4850 4860 4870 4880 4890 Wavelength(A) Wavelength(A)

7 20 CH Cygni CH Cygni 6 15 5 Resulting Resulting 4 10 3 2 5 relative intensity CH Cygni June 4 2006 relative intensity 1 CH Cygni June 4 2006 0 M6 III (RZ Ari) 0 M6 III (RZ Ari) 6540 6550 6560 6570 6580 4840 4850 4860 4870 4880 4890 Wavelength(A) Wavelength(A)

Fig. 1.— The spectra around Hα on October 13, 2005 Fig. 2.— The spectra around Hβ on October 13, 2005 (the (the upper panel), and on June 4, 2006(the lower panel) upper panel), and on June 4, 2006 (the lower panel) after after the subtraction of the M-type spectrum. the subtraction of the M-type spectrum.

tra into all 28 dispersion orders. An observation journal III. SPECTRAL VARIATIONS is outlined in Table 1. The nominal spectral resolving powers, λ/∆λ, were The lower portion of each panel in Figs. 1 - 5 is 45000 for the 200 µm fiber and 30000 for the 300 µm explained in this section. It is noticed here that all fiber. The output dispersions covered a range from the spectra represented in Figs. 1 - 5 were made use 1.56 - 2.14 A˚ mm−1, relied on the wavelength. of in the normalized intensity scale. The units of the Data were reduced with the NOAO IRAF package abscissa in Figs. 1 - 5 are wavelengths, all values being involving bias, flat-field and spectra extraction pro- corrected to the sun. cesses. Both the pixel positions and the wavelengths In the present epochs, the Hα line was changed from were related to their 6th-order polynomials. a very weak single-peaked emission profile (in Octo- ber 2005) into a double-peaked line (in June 2006). Hβ showed a single-peaked emission profile during the present epochs, which corresponded with the redward emission component observed in April 2004, and the SPECTRAL FEATURES OF THE CH CYGNI 95

10 20 CH Cygni CH Cygni 8 15 Resulting

6 Resulting 10 4 CH Cygni Oct. 13 2005 5 relative intensity relative intensity CH Cygni Oct. 13 2005 2

0 M6 III (RZ Ari) 0 M6 III (RZ Ari) 4990 5000 5010 5020 5030 3860 3880 3900 3920 3940 Wavelength(A) Wavelength(A)

10 20 CH Cygni CH Cygni 8 15 Resulting

6 Resulting 10 4 5 relative intensity relative intensity CH Cygni June 4 2006 2 CH Cygni June 4 2006

0 M6 III (RZ Ari) 0 M6 III (RZ Ari) 4990 5000 5010 5020 5030 3860 3880 3900 3920 3940 Wavelength(A) Wavelength(A)

Fig. 3.— The spectra around [O III] 5007 A˚ on October Fig. 4.— The spectra around [Ne III] 3869 A˚ on October 13, 2005 (the upper panel), and on June 4, 2006 (the lower 13, 2005 (the upper panel), and on June 4, 2006 (the lower panel) after the subtraction of the M-type spectrum. panel) after the subtraction of the M-type spectrum. profile of Hβ in 2004 was stronger than Hα in October and 6584 A˚ were clearly observed in 2004, whereas they 2005. The H² line was very weak, although it might were not detected in 2005. In 2006, they had profiles have blended with Ca II H line. H8 and H9 lines ap- that were more enhanced than in 2004. peared to be very weak. The line profiles of Hα and [S II] 4069 A˚ and 6584 A˚ could be definitely con- Hβ in the present epochs are shown in the lower parts firmed in the spectra resulting from the subtraction of of each panel of Figs. 1 and 2. the spectrum from the M-type giant in October 2005 In early 2004, Hα and Hβ had double-peaked pro- and June 2006. This indicates that in these epochs, files, and in late 2004, they were altered into single- the [S II] matter surrounding CH Cygni was either not peaked profiles. On the contrary hand, in the 1977 out- expanding out or not flowing into the region of [S II] burst phase, they always showed double-peaked emis- formation. The line profiles of [O III] 5007 A,˚ [Ne III] sion profiles, and around the end of the outburst phase, 3869 A,˚ and [S II] 6584 A˚ are presented in the lower the blueward emission component became weak, with panels of Figs. 3 - 5, respectively. the absorption component around its line center be- ˚ coming gradually weakened and finally disappearing On June 4, 2006, the peak intensity of [O I] 6300 A (Yamashita and Maehara, 1979; Hack et al., 1986; Hack was stronger by about 2 times than that on October 13, et al., 1988). 2005. In 2004, its intensity had remained unchanged for one year. The main emission components of [O III] 4959 A˚ Ca II H and K lines had P Cygni profiles in October and 5007 A˚ were gradually more enhanced during the 2004 and in October 2005. However, in June 2006 these present epochs and turned more intense than those in lines had a single emission. Na I D lines had two emis- 2004. However, [O III] 4363 A˚ did not appear during sion components after October 2004 when the spectrum the present phases. of the underlying M-type giant was subtracted. Since ˚ Broad [Ne III] 3869 A appeared to present three 2004, their intensities continued to be increased. The components after 2004. [Ne III] and [O III] began to absorption line components of Ca II H and K lines, and appear around the end of the outburst phases (Hack et Na I D lines were already described in Paper II. al., 1988). Fe II 5018 A˚ lines were present around 5018 A˚ (Fig. After subtracting the spectrum of the M-type giant, 3). [Fe II] continued to remain intense from 2004 to ˚ which is described in the next section, [N II] 6445 A 2006. However, in the present epochs, fewer Fe II lines 96 YOO & YOON

7 3.0 CH Cygni 6 CH Cgyni H 2.5 5 Resulting 4 2.0 3 CH Cygni Oct. 13 2005 2 1.5 relative intensity 1

0 M6 III (RZ Ari) Intensity 1.0 4050 4060 4070 4080 4090 Wavelength(A) 0.5

7 CH Cygni 0.0 6 June. 4 2006 5 Resulting -0.5 -400 -200 0 200 400 4 Velocity (km/s) 3 CH Cygni June 4 2006 2 Fig. 6.— Gaussian deconvolution for the resultant Hα on relative intensity June 4, 2006. The solid line denotes the observed line. The 1 bold solid, the solid, the dashed, the dotted, and the dash 0 M6 III (RZ Ari) dotted lines mean the result of fitting and each Gaussian 4050 4060 4070 4080 4090 component. Comparison of Gaussian profiles and a syn- Wavelength(A) thetic rectangle-like line profile would be noticed. The red line profile representing a theoretical model for the accre- Fig. 5.— The spectra around [S II] 4069 A˚ on October tion disk is over-plotted. 13, 2005 (the upper panel), and on June 4, 2006 (the lower panel) after the subtraction of the M-type spectrum. and Na I D had double absorption cores. were identified than [Fe II]. Moreover, the Fe II line IV. SUBTRACTION OF THE UNDERLY- profiles were asymmetric. ING M-TYPE SPECTRUM Many TiO bands were observed in the strong absorption spectra. Since 2004, the depth of the (4, 0) On October 13, 2005, and June 4, 2006, CH Cygni band head of the α-system of TiO at 4424 A˚ in these spectra were compounded with emission lines and the phases became gradually shallow, but it was still alive. spectrum of the M-type giant. To detect the actual The depth of the TiO band head at 4424 A˚ in October emission lines, removal of the underlying spectrum of 2004 was deeper by about 2 times than that in October the M-type giant from CH Cygni was carried out simi- 2005, and the depth in October 2005 was comparable lar to that proposed in Paper II. The HD18191 (RZ Ari- to that in June 2006. etis, M6 III) spectrum was used as the standard. Since 2004, the depth of the Ca I 6573 A˚ line of the The resultant line profiles of Hα are shown in the M-type giant became gradually weak, whereas that of upper portion of Fig. 1. The resulting profile of the Hα the Sr II 4215 A˚ line remained nearly unchanged. emission lines might have at least four emission components and no absorption components as in October In 2004 - 2006, behaviors of spectral lines of CH 2005 and June 2006. The intensities of the Hα emis- Cygni is summarized as follows. Hα showed a single- peaked emission line in October 2005, which subse- sion lines on October 13, 2005, and June 4, 2006 are quently changed into a double-peaked profile in June evidently less intense by about two times than and sim- 2006, with no absorption line observed at its line- ilar to those in October 2004, respectively. The widths center. In April 2004, Hα and Hβ showed double- of the Hα emission lines on October 13, 2005, and June peaked emission profiles with an absorption line around 4, 2006, are less than those in 2004. It implies that the each of their line-centers. In the present epochs, nebu- clouds forming Hα in October 2005 evidently changed lar emission lines of [O III] and [Ne III] were produced; to a smaller area than those forming Hα in October in addition, more Fe II and [Fe II] lines were observed 2004. than in 2004 Paper I; Paper II . Ti II and Cr II ob- The resulting line profiles for Hβ 4861 A,˚ [O III] 5007A,˚ served during the past outburst phases were absent in [Ne III] 3869 A,˚ and [S II] 4069 A˚ are obtained in the the present epochs. Moreover, in the present epochs, same manner as mentioned above and are provided in emission lines of He I, [O I] and [O III] were observed the upper panels in Figs. 2 - 5. They do not appear SPECTRAL FEATURES OF THE CH CYGNI 97

6 5 CH Cgyni [O III] CH Cgyni [Ne III] 5 4

4 3

3 2 Intensity Intensity 2

1 1

0 0 June 4 2006 Oct. 13 2005

-400 -200 0 200 400 -400 -200 0 200 400 Velocity (km/s) Velocity (km/s)

Fig. 7.— Gaussian deconvolution for the resultant [O III] Fig. 8.— Gaussian deconvolution for the resultant [Ne III] 5007 A˚ on June 4, 2006. The solid line denotes the observed 3869 A˚ on October 13, 2005. The solid line denotes the line. The thin solid, the dashed, the dotted, the solid, and observed line. The bold solid, the solid, the dotted, and the dash dotted lines mean the result of fitting and each the dash dotted lines mean the result of fitting and each Gaussian component. Gaussian component. to be greatly affected even after the subtraction of the velocity. Therefore, in this study, for the sake of sim- spectrum of the M-type giant. H8 and Ca II K lines plicity, a likelihood model is assumed as a Gaussian are observed around 3889 A˚ and 3933 A,˚ respectively. function, and then the Gaussian deconvolution is car- The resultant [O III] 5007 A˚ had broad emission ried out. components and its intensity was more enhanced on The emission lines are believed to be optically thin. June 4, 2006 than that on October 13, 2005. Hence, if the velocity distribution of gas in the cloudlet The intensity of the strongest emission component is assumed as a Gaussian function, flux density emitted of [Ne III] 3869 A˚ in June 2006 was decreased by half from the i-th cloudlet is represented by than that on October 13, 2005. However, the intensity of the strongest emission hν component of [S II] 4069 A˚ on October 13, 2005, was Fi = neiniα Vi φi (1) weaker by about 10 % than that on June 4, 2006. The 4π profiles of [S II] appeared to be asymmetric. [S II] λ − λ0i 2 = F0i exp(−( ) ), 4076 A˚ was clearly observed in the resulting subtrac- σi tion spectra. 1 λ−λ0i 2 vri where φi = √ exp(−( ) ), λ0i = λ0(1 + ), Intensities of the emission lines of the resultant πσ σi c Na I D and D profiles became gradually stronger after hν √1 1 2 F0i = neiniα 4π Vi πσ , λ0 is the wavelength at the cen- 2004. ter of the function, ne an electron density, n an ion density, α an recombination coefficient, h the Planck V. GAUSSIAN FITTING constant, ν a frequency of the line, Vi a fractional part i of the emitting volume, and σ a standard deviation, In section 3, the spectral line variations of CH Cygni and v the radial velocity of the cloudlet i. were described in detail. All the emission lines involv- ri ing blue continua in these epochs were stronger than If there are n cloudlets in the cloud, total flux density those in 2004. is given by Because all the emission lines observed during these observation epochs had complicated profiles with sev- F = ΣF (2) eral sub-peaks, each component of an emission line, i when decomposed, might be arisen from a cloudlet in λ − λ0i 2 = ΣF0i exp(−( ) ). a gas cloud. The cloudlet might be expanding into the σi circumference space around the system with its radial Here the sum of i is the summation of n cloudlets. 98 YOO & YOON

peak intensity of the Gaussian function from a local 5 continuum; Vr denotes the radial velocity; and VD is CH Cgyni [S II] the Doppler velocity of each Gaussian component. 4 During the present epochs, the Doppler velocities of the broad components, Ebr, of Hα on October 13, 2005, and on June 4, 2006 were smaller than those of the 3 respective components in October 2004. In addition, the intensity of the strongest component, Em, of Hα was smaller by about 0.4 times on October 13, 2005, 2 and was larger by about 1.1 times on June 4, 2006, Intensity than that of the corresponding line in October 2004. The resultant synthetic line profile is well consistent 1 with the simplified likelihood model of the Gaussian functions. It is a reason why the observed line profile, the most fiducial line profile, must be based on even 0 Oct 13 2005 in a comparison with any of many theoretical models. However, the broad emission wings of the observed Hα -400 -300 -200 -100 0 100 200 line might be the sum of several of such a rectangle Velocity (km/s) profile. It means that the accretion disk might consist of several ring structures. Fig. 9.— Gaussian deconvolution for the resultant [S II] The Doppler velocities of the broad components, ˚ 4069 A on October 13, 2005. The solid line denotes the Ebr, of [O III] 5007 A˚ and [Ne III] 3869 A˚ on Octo- observed line. The bold solid, the solid, the dotted, and ber 13, 2005 and on June 4, 2006 became larger than the dash dotted lines mean the result of fitting and each Gaussian component. those of the respective components in October 2004. Intensities of the strongest component, Em, of [O III] 5007 A˚ and [Ne III] 3869 A˚ on October 13, 2005, and It is uncertain that all the emission lines could be on June 4, 2006 became lower than those of the same decomposed by Gaussian components. For example, of in October 2004. This implies that these elements the broad emission component of the [O III] nebular were placed next to each other, and were kinematically line might correspond with a rectangular line profile affected by each other. that the expanding shell structure produces. There- On October 13, 2005, and on June 4, 2006, the fore, it is believed that each of Gaussian components Doppler velocities of the broad components, Ebr, of might be formed from either the expanding gas shell [N II], and [S II] were about two hundred km s−1. or a fine structure of the cloudlet. And, the double- peaked profiles might be originated from the edge-on VI. RADIAL VELOCITIES accretion disk (Robinson et al. 1994). The Hα and Hβ lines observed during these observation epochs had After the subtraction of the M-type spectrum, the such profiles. They showed also complicated profiles line center could not be exactly decided even for the with several sub-peaks, which might be originated from line peak of the strongest emission component. The either a portion of the accretion disk or independently radial velocities were measured around the center of moving-out matter from the hot star. the line peak at about 50 % - 90 % of line intensi- Therefore, to investigate how these emission lines ties. Hence, there were small differences in the radial were formed, the observed emission lines were decon- velocities between the observed lines and the Gaussian voluted with several Gaussian functions. The Gaus- emission components for the referred lines, which range −1 sian function is adopted from the equation of Ikeda and from 1 - 3 km s . Discrepancies in the widths of the Tamura (2004). A program used for de-convolution of small sharp humps in both observed and resulting line the observed lines into Gaussian functions was written profiles might be more than about 0.07 A.˚ by Dr. J.C. Chae. The radial velocities of the emission lines of H I, Examples of the Gaussian deconvoluted line pro- [O I], [O III], [Ne III], [Ca II], [N II], and [S II] were files of the resulting Hα and [O III] 5007 A˚ spectra on measured. Tables 3, 4 and 5 present the radial veloc- June 4, 2006, and of the resulting [Ne III] 3869 A˚ and ities measured for emission lines, forbidden lines, and absorption components, respectively. All the velocities [S II] 4069 A˚ spectra on October 13, 2005 are presented are corrected to the Sun. When several lines were mea- in Figs.6-9 , respectively. Velocities in the abscissa of sured, the probable errors are quoted. Figs. 6 - 9 are corrected to the sun. The parameters of the Gaussian function are listed in Table 2. In Table 2, λ0 is the wavelength at the center of the deconvoluted Gaussian function; I is the SPECTRAL FEATURES OF THE CH CYGNI 99

Table 2. Parameters of the Gaussian components for the element of CH Cygni Date Element wavelength Component λ0 I FWHM Vr VD (A)˚ (A)˚ (A)˚ (km s−1) (km s−1) Oct. 13, 2005 Hα 6562.817 Ebr 6561.69 0.34 3.42 -51.8 188.5 Em 6561.69 0.72 0.58 -51.8 31.9 E1 6562.12 0.60 0.57 -32.0 31.4

June 04, 2006 E1 6560.67 0.31 1.00 -98.5 55.0 Ebr 6561.64 0.61 4.15 -53.7 228.4 Em 6561.78 1.64 0.59 -47.7 22.5 E2 6562.74 0.16 1.14 -3.8 63.0

Oct. 13, 2005 Hβ 4961.332 Ebr 4860.48 1.20 3.42 -52.6 254.5 Em 4860.48 5.10 0.80 -52.6 59.4 E1 4860.82 0.21 0.35 -31.4 26.2

June 04, 2006 E1 4859.75 0.78 0.97 -97.6 71.8 Ebr 4860.48 0.65 3.85 -52.3 286.1 Em 4860.54 3.02 0.60 -48.8 44.6 E2 4861.28 1.11 0.99 -3.5 73.6

Oct. 13, 2005 [O III] 5006.84 E1 5005.30 0.97 0.70 -92.5 50.6 Ebr 5005.87 1.04 4.47 -58.4 322.6 Em 5005.94 1.95 0.62 -53.6 44.7 E2 5006.52 1.52 0.69 -19.5 50.1

June 04, 2006 E1 5005.52 1.77 1.76 -78.8 127.4 Ebr 5005.98 0.74 4.86 -51.7 350.6 Em 5006.00 3.45 0.47 -50.6 33.9 E2 5006.65 0.98 0.47 -11.1 33.9 E3 5007.48 0.63 1.40 38.5 100.8

Oct. 13, 2005 [Ne III] 3868.74 Ebr 3867.97 1.24 3.00 -65.6 280.6 Em 3868.13 2.05 0.66 -47.2 61.3 E1 3868.67 0.62 0.47 -5.4 43.9

June 04, 2006 Ebr 3867.94 0.54 2.02 -62.4 188.5 Em 3868.03 1.05 0.47 -55.1 43.9 E1 3868.44 0.28 0.43 -23.0 39.8

Oct. 13, 2005 Na I D 5889.953 Ab1 5887.57 0.66 0.45 -121.2 27.8 Ab2 5887.97 0.62 0.44 -101.0 26.9 Ab3 5888.46 0.54 0.16 -76.0 10.0 E1 5888.84 0.35 0.33 -56.7 20.3 E2 5889.15 0.22 0.30 -40.5 18.3

5895.923 Ab1 5893.53 0.44 0.24 -122.0 15.0 Ab2 5893.83 0.94 0.37 -106.7 22.9 Ab3 5894.41 0.67 0.18 -76.8 11.1 E1 5894.78 0.27 0.29 -58.3 17.5 E2 5895.16 0.22 0.33 -39.0 20.3

June 04, 2006 Na I D 5889.953 Ab1 5887.59 0.54 0.29 -120.4 18.0 Ab2 5887.91 0.88 0.34 -104.1 21.0 Ab3 5888.30 0.51 0.14 -83.9 8.4 E1 5888.66 0.19 0.39 -65.7 24.0 E2 5889.02 0.11 0.25 -47.5 15.2

5895.923 Ab1 5893.72 0.74 0.34 -111.9 20.9 Ab2 5893.95 0.44 0.39 -100.3 23.9 Ab3 5894.31 0.61 0.14 -82.2 8.4 E1 5894.63 0.17 0.39 -65.9 23.9 E2 5895.00 0.11 0.25 -46.7 15.2

Oct. 13, 2005 Ca II 3933.664 Em 3932.09 1.01 0.15 -120.4 14.1 Ebr 3933.02 2.02 0.38 -49.2 34.7

June 04, 2005 Em 3932.91 1.23 0.39 -57.3 36.2

Oct. 13, 2005 [S II] 4068.62 Ebr 4067.79 1.22 2.07 -60.8 183.9 Em 4067.70 1.07 0.50 -67.9 44.1 E1 4068.32 0.35 0.40 -22.3 35.1

June 04, 2006 Ebr 4067.88 0.67 2.24 -54.6 199.0 Em 4067.71 0.98 0.42 -67.1 37.0 E1 4068.24 0.47 0.38 -28.2 33.9

June 04, 2006 [Ne III] 6583.6 Em 6582.22 0.08 1.12 -62.7 61.4 Ebr 6582.17 0.11 3.79 -65.3 208.2 E1, E2, E3: Emission components. Ab1, Ab2, Ab3: Absorption components. Ebr: The broadest emission component. Em: The strongest emission component. 100 YOO & YOON

Table 3. Radial velocities(km s−1)measured for observed emission lines Date Hα Hβ Hγ Ca II H & K Fe II (Number of lines) Oct. 13, 2005 -51.3 -52.7 -52.6 -41.7±1.6 -67.3±1.9 (3) June 4, 2006 -107.0, -44.1 -51.6 -55.7 -47.7±1.6 -65.1 (1)

Table 4. Radial velocities(km s−1)measured for forbidden lines Date [O I] [O III] [S II] [N II] [Ne III] [Fe II] (Number of lines) 6300A˚ 4959,5007A˚ 4069A˚ 6582A˚ 3869A˚ Oct. 13, 2005 -65.2 -52.3±0.4 -69.0 -49.7±1.9 -65.5±2.1 (4) June 4, 2006 -66.6 -52.2±0.6 -68.3 -56.7 -56.7 -71.6±2.1 (4)

VII. DISCUSSION AND CONCLUSION From these observations, the story regarding the evolution of CH Cygni is hypothesized: in April 2004, The model involving an accretion disk and jets is an accretion disk around the hot star was heading in introduced to interpret the characteristics of each of the direction of its destruction; in October 2004 and the lines identified in October 2005 and June 2006. October 2005, the jets were out-flowing; in June 2006, After 2004, the intensities and the line profiles of Hα the accretion disk adopted a destruction mode. continuously changed. For example, for Hα, the V/R Results of decomposing the Hα line in 2006 are was < 1 in April 2004, << 1 in October 2004, < 1 sketched. The blue component, E1, at 6560.7 A˚ and in October 2005, and ∼ 0.1 - 0.2 in June 2006. Here, the strongest one, Em, at 6561.8 A˚ are originated from V and R represent the intensities of the blueward and the accretion disk. The another one, E2, might be a redward peaks of Hα. In October 2005, the Hα and Hβ fine structure in the disk. lines had single-peaked emission lines, implying the ex- During the past eruption phases, the blue continuum istence of out-flowing radio jets. And the Hα line was was believed to be produced due to the accretion disk weaker than the Hβ line. In June 2006, their intensities (Yoo and Yamashita, 1984). However, Contini et al. were opposed to each other. Conspicuous variations in (2008) suggested the UV lines were formed in expand- the H I line profiles of CH Cygni from October 2004 ing shocked nebulae. to June 2006 might result from the collapse of the accretion disk and its subsequent formation. On the con- According to Shakura and Sunyaev (1973), the coef- trary, intensities of Hβ lines in 1983, one year before ficient of kinematic viscosity ν is represented as αCsH. the 1984 eclipse on CH Cygni, were weaker than those Here, α is a viscosity parameter, Cs the sound velocity, and H a scale height of the disk in the z-direction. in 1984, the beginning epoch of the eclipse (Hack et al., 2 And, a viscous time scale is R ∼ R . R is the radius of 1988). ν Vr In general, in October 2005 and June 2006, the the disk. If using radial velocities of the emission com- shapes of the lines in the resultant Hα were widely dif- ponents of the Hα line and a standard α-disk model, H ferent. |Vr| ∼ αCs R > 1. Since 0 < α < 1, H > R. This As shown in Fig. 10, from April 2004 to Octo- indicates the accretion disk was newly born. In case ber 2004, the peak intensity, I, of the strongest emis- an accretion disk was newly formed, it is expected to sion components, Em, of the resulting Hα increased, have been born between October 2005 (φ = 0.52) and whereas from October 2004 to October 2005, the I of June 2006 (φ = 0.57). Assume that only emission com- Em of the resulting Hα decreased, and after those two ponents E1 and Em of the Hα line consist of a new epochs, the I of Em of the resulting Hα increased. resulting profile of a Hα line. If so was it, FWHM of the resulting profile of a Hα line might be correspon- As shown in Fig. 11, the Doppler velocities, VD, of dent to the full separation of the double peaks of the the broad emission components, Ebr, of the resulting Hα. Since FWHM = Vo sin i, the outer radius of the Hα steeply increased from April 2004 to October 2004, 8 accretion disk, ro, is approximately 7 × 10 km, which whereas from October 2004 to October 2005, these is within the Lagrangian point. Here, the inclination Doppler velocities, VD, slowly decreased, and subse- of the accretion disk, i, is adopted by 80o (Robinson et quently remained nearly unchanged. In October 2005, −1 al. 1994; Skopal 1995). From the radial velocities of 500 km s of the Doppler velocity of H I was above E1 and Em of the Hα, assume that an orbital velocity the shock speed. of an inner region of the accretion disk was about 25 SPECTRAL FEATURES OF THE CH CYGNI 101

Table 5. Radial velocities(km s−1)measured for absorption lines Date Ca II Na I M-type (Number of lines) H&KD1 and D2 Oct. 13, 2005 -110.9±0.1 -108.8±2.9 -75.7±0.4 -59.6±0.9 (11) June 4, 2006 -110.9±1.0 -79.3±1.0 -50.4±0.6 (10)

Table 6. Normalized intensities and Doppler velocities of the resultant Hα, [O III] and [Ne III] in April 9, 2004 and October 2, 2004 −1 IVD(km sec ) Apr. 9, 2004 Oct. 2, 2004 Apr. 9, 2004 Oct. 2, 2004 Hα 0.2 1.1 262 502 [O III] 5007A˚ 2.9 8.4 100 148 [Ne III] − 1.2 − 56

All data are adopted from Figs 8 and 10 of Yoo (2007), and Tables 1, 2 and 3 of Yoo (2007). − means out of range of the spectrum in April, 2004.

km s−1. From this, its inner radius could be estimated of the resulting [O III] line appeared to have rapidly as 10 % - 20 % of its outer radius. increased, and the peak intensity of its strongest emis- The resulting Ca II H and K lines in October, of sion component also gradually increased. It indicates both years 2004 and 2005, showed the P Cygni pro- that the cloudlets emitting the [O III] lines might be files, whereas that in June 2006 did not present the moved out by the out-flowing radio jets. For exam- characteristic profile. The emission components of the ple, the broadest emission component, Ebr, and the resultant Ca II H and K lines were observed when Hα sharper main emission component, Em, of [O III] 5007 had a single-peaked profile. Hence, the emission com- A˚ in June 2006 were formed by the jets in October 2005 ponents of the resultant Ca II lines might have origi- and by jets in October 2004, respectively. The sharper nated in the low density region around the out-flowing main emission component, Em, of [O III] 5007 A˚ in radio jets. June 2006 might be due to the collimated jet streams The radial velocity of the [O I] 6300 A˚ was used to in October 2004. In this context, AG Pegasi showed arrive at the systemic velocity (Wallerstein, 1983). Its differences in the radial velocities of many of the per- radial velocities did not appear to have changed in 2004. mitted lines due to the occurrence of the shock waves Nevertheless, on October 13, 2005, and on June 4, 2006, (Yoo, 2008). changes in the [O I] line profiles were observed. This The resultant [Ne III] lines appeared to have at least implies that even the circumference cloud that formed three emission components. As shown in Fig. 10 and the [O I] lines was symmetric in 2004, being affected 11, from April 2004 to October 2005, the Doppler ve- by the radio jets that arose in both October 2004 and locities, VD, of the broad emission components of the October 2005. resulting [Ne III] and the resulting [O III] were grad- Using Tables 2 and 6 in this study, In Table 6,line ually increased, whereas from October 2005 to June intensities and expansion velocities of Gaussian compo- 2006, that of [Ne III] varied with a gradient opposite nents of the Hα, [Ne III] and [O III] lines with respect to that of [O III]. This implies that the [Ne III] clouds to the observation dates are plotted in Figs. 10 and took the important role of transporting their momenta 11. I presents the peak intensity of the Gaussian com- to the [O III] clouds. The decrease in the velocities of the [Ne III] clouds results in a little increase in their ponent in unites of a local continuum, and VD denotes the Doppler velocity of the Gaussian component. optical thickness during the period from October 2005 to June 2006. The [Ne III] clouds helped the strongest As shown in Figs. 10 and 11, in the present epochs, ˚ the resulting [O III] lines might have more than four emission component, Em, of the [O III] 5007 A to be- emission Gaussian components. [O III] lines were come even sharper. From these discussions, the clouds expected to be formed in the several circumference forming the [Ne III] line might be concluded to be lo- cloudlets around the hot star of CH Cygni. Just af- cated inside the clouds emitting the [O III] line. ter 2004, the Doppler velocity of the broad component The [S II] line profiles were asymmetric and had 102 YOO & YOON

10 600 CH Cygni CH Cygni

500 8 ) -1 400 6

300

4 200 Normalized Intensity Doppler Veocity(km s

2 100

0 0

2004.4 2004.10 2005.10 2006.6 2004.4 2004.10 2005.10 2006.6 Date Date

Fig. 10.— Normalrized intensities in units of local con- Fig. 11.— Doppler velocities of resulting broad Gaussian tinuum of resulting Gaussian functions of Hα, [O III] and functions of Hα, [O III] and [Ne III] with the dates. [Ne III] with the dates. Weak intensity components are excluded. dius on October 13, 2005 than on June 4, 2006. Around October 13, 2005, the stellar matter of the M III star more than three emission components. This implies was transferred to the hot star, forming an accretion that the [S II] clouds originated in the circumference disk around the hot star on around June 4, 2006. How- material at the outskirts of an ellipsoidal envelope that ever, no observational data is available as to when the was formed as a result of the out-flowing radio jets. accretion disk that appeared on October 13, 2005, col- The emission components of the resulting Na I D lapsed. The Hα lines of CH Cygni in April 2004 had lines became enhanced from 2004 to the present epochs. double emission peaks, characteristic of an accretion Their radial velocities were in agreement with those of disk. On October 13, 2005, and on June 4, 2006, many Hα. Hence, Na I D and Ca II H and K emission compo- elements within the circumference of CH Cygni might nents might be also formed because of the out-flowing have been on head-on collision courses. radio jets. The absorption components of Na I D, −1 whose radial velocities were about -77 km sec , might ACKNOWLEDGEMENTS have their origin in the M-type star. It is well known that a leading cause for the forma- The authors would be grateful to Professor Ya- tion of Na I D lines is spin-orbital interactions of the mashita for his helpful comments of improving this electrons of Na I in 1.84 × 105 G of the strong mag- work, and also thank the referee for some comments netic field around the hot star of CH Cygni. Assum- that improved this paper. We acknowledge with tha- ing the surface magnetic field of a hot star was about nks the variable star observations from the AAVSO 107 G, and the hot star had a dipole magnetic field, the International Database provided the light curve of magnetic field falls off as B ∼ B ( R∗ )3, where R , the CH Cygni. The authors appreciate all the help and ∗ r ∗ kindness of BOAO staff and observation operators dur- radius of the hot star, is the 0.01R¯, and B∗ its surface magnetic field. Hence, the region forming the Na I D ing their stay for observations. lines is placed approximately at 2.64 × 104 km, which is unexpectedly close to the hot star. Assuming this REFERENCES distance to be an Alfven radius and the spherical symmetric radial inflow of matter toward the hot star, the Contini, M., Angeloni, R., & Rafanelli, P., 2008, The accretion rate needed forming the Na I D lines could be symbiotic star CH Cygni. I. An analysis of the −6 adopted as approximately 10 M¯/ year, which are shocked nebulae at different epochs, arXiv 0807, larger by about 10 times than the estimated accretion 1480 rate of the upper limit. Eyres, S. P. S., Bode, M. F., Skopal, A., Crocker, A brief scenario in this study is as follows. According M. M., Davis, R. J., Taylor, A. R., Teodorani, M., to the AAVSO, CH Cygni was fainter on October 13, Errico, L., Vittone, A. A. & Elkin, V. G., 2002, The 2005 than on June 4, 2006. The pulsating M III star in symbiotic star CH Cygni-II. The ejecta from 1998- the the companion star of CH Cygni had a larger ra- 2000 active phase, MNRAS, 335, 526 SPECTRAL FEATURES OF THE CH CYGNI 103

Faraggiana, R. & Hack, M., 1969, Spectroscopic evi- dence for mass loss from CH Cgni, AP&SS, 3, 205 Hack, M., Engin, S., Rusconi, L., Sedmak, G., Yilmaz, N., & Boehm, C., 1988, Line spectrum variations during the CH Cygni outburst of 1977-1986, A&AS, 72, 391 Hack, M., Rusconi, L., Sedmak, G., Aydin, C., Engin, S., & Yilmaz, N., 1986, CH Cygni-The outburst of 1977-85, A&A, 159, 117 Hinkle, K. H., Fekel, F. C., Johnson, D. S., & Scharlach, W. W. G., 1993, The triple symbiotic system CH Cygni, AJ, 105, 1074 Ikeda, Y. & Tamura, S., 2004, Spectroscopic Diagnos- tics of Symbiotic Stars II. Statistical Analyses of Highly Resolved Emission Line Proﬁles, PASJ, 56, 353 Mikolajewski, M., Mikolajewska, J., & Khudiakova, T. N., 1990, The long-period symbiotic binary CH Cygni, A&A, 235, 219 Robinson, K., Bode, M. F., Skopal, A., Ivison, R. J., & Meaburn, J., 1994, On the nature of the emission- line proﬁles of symbiotic stars-I. Accretion discs, MNRAS, 269, 1 Shakura, N. I. & Sunyaev, R. A., 1973, Black holes in binary systems. Observational appearance, A&A, 24, 337 Skopal, A., Bode, M. F., Crocker, M. M., Drechsel, H., Eyres, S. P. S., & Komzik, R., 2002, The symbiotic star CH Cygni-IV. Basic kinematics of the circumstellar matter during active phases, MNRAS, 335, 1109 Skopal, A., 1995, Phtometric observations of eclipes in the symbiotic triple system CH Cygni, IBVS, 4157, 1 Wallerstein, G., 1983, CH Cygni 1981-82 - Now a shell star, PASP, 95, 135 Yamashita, Y., 1967, MK Spectral Types of Bright M- Type Stars, PDAO, 13, 47 Yamashita, Y. & Maehara, H., 1979, A Binary Model for CH Cygni, PASJ, 31, 307 Yoo, K. H. & Yamashita Y., 1984, Radial velocity components and their variations in the circumstellar spectrum of CH Cygni, PASJ, 36, 567 Yoo, K. H., 2006, The high resolution spectra of symbiotic star CH Cygni in 2004, NewA, 11, 359, Paper I Yoo, K. H., 2007, CH Cygni in 2004 II: Nebular spectrum features, NewA, 12, 569, Paper II Yoo, K. H., 2008, High resolution spectra of the symbiotic star AG Pegasi in 2004-2006, NewA, 13, 646